Research question/goal:

Recent coalition theories assume that parties are both office- und policy-oriented. Equilibrium solutions of coalition negotiations are derived non-cooperatively via backwards induction assuming a fixed sequence of negotiation steps which are started by the choice of a formateur. But these sophisticated models are tested rarely. The main goal of the project was to study the effects of office and policy motivations of German parties on coalition building in the German states and to develop alternatives to formateur models which are based on assumptions not holding for Germany.
For that purpose a new data basis on the policy positions and ministerial ambitions of German state parties had to be developed. We relied on election manifestos for state elections from 1975 to 2010 to derive both measures of interest intensity and policy positions in various policy domains. The section headings of the platforms were coded as belonging to a certain policy domain, the text length dedicated to a domain was interpreted as a salience measure and the domain specific policy positions were extracted from the text by Wordfish.
Government formations within the parliamentary systems of the German states are an ideal test ground for coalition theories. The researcher is able to focus on the theoretically central negotiation topics of ministerial assignments and policy compromises without much interference of institutional and party system peculiarities as in international comparisons. One is able to capitalize on a most similar systems design. Bargaining is “freestyle” without formateurs who would have to be chosen by a non-existent head of state. One result of the project is that the largest party will lead the new government if it dominates the coalition game in the sense that it has numerically more opportunities to form minimal winning coalitions than every other party and that the non-dominant largest party or the second largest party will become the party of the prime minister depending on which one has the shortest distance to the centroid of the multidimensional policy-domain specific space (Pappi/Seher).
Shikano and Linhart applied a solution concept of cooperative game theory (according to Sened) to estimate the relative strength of office vs. policy motivations of the individual German parties. They found quite surprisingly that Christian Democrats are more policy motivated and Social Democrats are more office motivated when bargaining new coalitions. The average policy distance between Christian and Free Democrats is smaller when they form a coalition than when they not form one. Policy distances among the parties on the left are usually larger than on the right and do not vary with coalition building. Policy distances between Christian and Social Democrats are sometimes small and avoiding a grand coalition in spite of that indicates a strong office motivation.